Saturday, March 28, 2009

Readership vs Viewership

In which a new and more accurate metric for comparing blog readerships is proposed. I was intending to link to a rather amusing post on Pharyngula - I'll get around to it, as the humor inherent in science fetishists moaning about science funding is too good to pass up - when this exchange between commenters there set me to pondering the unreliability of visit statistics.

#443: I haven't learned too much about science in the time I've spent reading this blog. PZs posts boil down to:

1) The latest creationist attacks "threatening" science
1.5) Calling these people and those with faith idiots
2) Look-who-hates-me comments
3)Pictures of invertebrates

No, can't say I've learned a lot about science from here. Based on PZs posts and the comments of those here constantly on his jock, I have learned, however, that science is pervaded by dicks. And that isn't cool. People tend to paint with a broad brush, and I don't want to be lumped in with you fools. Luckily this blog has fewer readers than a class of preschoolers, so I think I'm safe."

#444: "Pharyngula is a very popular blog and with your post, you are now on Google. The average number of views per day is 75,000+. We don't treat the stupid nicely, and that appears to include you. The only dick here is you."

#455: "Very popular blog, eh? 75,000+ views? Look, there are about 500 solid readers of Pharyngula, and each hits Pharyngula about 150 times per day to see if someone has replied to his/her comment in a thread. You'll probably shoot back that the users are unique based on IP address or some other nonsense, but I don't care...."

The problem in attempting to assess readership by visits or page views is that, as the commenter pointed out, the "unique visitor" concept does not pass logical muster. Consider that the average reader reads 200 words per minute, which is 3.33 words per second. Since the average visit length to Pharyngula is only 9 seconds, this means that the average read per visit is 30 words. The post that the commenters were discussing is 920 words, which indicates that either 1) Pharyngulans needed to visit his blog more than 30 times on average in order to read the whole thing, 2) Pharyngulans are all extraordinary speed readers who read several times faster than a very fast speed reader's rate of 16.67 words per second, or, rather more likely, 3) Pharyngula is far less popular than the Sitemeter statistic of average visits per day would indicate and the number of visits are, as the commenter surmised, being significantly inflated by recounting regulars who bang the refresh bar like trained monkeys in order to read new comments.* This isn't just a Pharyngula phenomenon either, as many of the supposedly more popular blogs show the same sign of visit-count inflation.

Compare, for example, Pharyngula with a blog that is theoretically the same "size", but is noticeably more influential in the blogosphere and theoretically more widely read, Ace of Spades HQ.

V/D: 72,120
VL: 9 seconds
RPV: 30 words
RPD: 2,163,600 words

V/D: 72,709
VL: 16 seconds
RPV: 53 words
RPD: 3,873,936 words

It's clear that although the number of visits per day are almost identical, people read nearly twice as much at Ace's as they do at PZ's. Now, comparing the words Read Per Day at three of the biggest blogs by conventional visits/day indicates another interesting factor, the division between content creation and traffic direction.

V/D: 252,763
VL: 5 seconds
RPV: 16.65 words
RPD: 4,208,503 words

V/D: 380,532
VL: 1 second
RPV: 3.33 words
RPD: 1,267,172 words

V/D: 726,387
VL: 1 second
RPV: 3.33 words
RPD: 2,418,869 words

Obviously, Instapundit is much more valued for the aggregation aspect of his posts than anything he's writing, which makes sense because he isn't actually writing much on that site; his value derives from his role as a traffic director towards the content created by others. In other words, he, Drudge, and other traffic directors serve the Gatekeeper role that the mainstream media once played in addition to creating actual content. Daily Kos - at least, the home page - appears to serve much the same role, whereas at least a small percentage of Michelle Malkin's visitors spend some time actually reading what she writes; she possesses an actual readership as well as a viewership and acts in both capacities. I use the term "viewership" because it's obviously absurd to speak of a "readership" that reads less than four words per visit! Now, here's where it gets interesting. I selected three blogs that I know people read rather than use as primary traffic pointers.

V/D: 23,741
VL: 200 seconds
RPV: 666 words
RPD: 15,811,506 words

V/D: 5,750
VL: 359 seconds
RPV: 1,195 words
RPD: 6,873,953 words

V/D: 4,380
VL: 105 seconds
RPV: 350 words
RPD: 1,531,467 words

All hail Althouse! Ann Althouse has a great visit length in addition to a large visitor count and proves to have a much larger readership factor than I would have guessed. Dr. Helen's visit length is astounding, giving her excellent RPV numbers, although that didn't surprise me quite as much because her posts often stimulate long and thoughtful comment threads that indicate a lot of people are actually reading them. Rachel Lucas came out well and about how I expected given her amusing posts and a reliable number of regular commenters discussing what she wrote about. My blog, on the other hand, didn't do quite as well in comparison, which I found a little surprising in that this blog is usually considered by other bloggers to be more discussion-heavy than the norm. Nevertheless, the RPV readership metric tends to indicate that blogs like this one and Rachel Lucas's have readerships that are similar in size to Pharyngula's rather than being one-fourteenth the size as the visits per day would have it.

V/D: 5,143
VL: 136 seconds
RPV: 452 words
RPD: 2,324,636 words

I should note that one potential flaw in the RPD metric is that I don't know how embedded video affects the visit length, although I haven't seen any change in VP's average VL as a result of that. I should probably check out the VL on a number of sites that make heavy use of embedded video, although it occurs to me that all this is likely to prove is that the Internet is mostly used for goat porn or something. But at least the RPD statistic does show the inherent flaw in using visits as a means of comparing one blog's readership with another's. Even so, a stat that divides RPD numbers by one million is probably a more meaningful comparison of actual blog readerships than supposedly unique visitor stats.

*Ann Althouse points out that there is a fourth possibility, which is that the Visit Length statistic is even less reliable than the Unique Visits statistic. Which, of course, is true.


Post a Comment

Rules of the blog
Please do not comment as "Anonymous". Comments by "Anonymous" will be spammed.

<< Home

Newer Posts Older Posts